This study examines how prior knowledge influences future learning. Although it has been established that prior knowledge does have a strong impact on later learning, it is not clear how this effect manifests itself. Previous research suggests that one way to examine the effects of knowledge on learning is to compare subtasking, that is, dividing a task into subsets, to unstructured learning, or learning "all-at-once". Neural network simulations using the cascade-correlation learning algorithm predict that subtasking facilitates learning when it involves learning a function that is compatible (i.e., logically consistent) with the rest of the task, and hinders learning when it involves learning a function that is incompatible with the rest of the task (Tetewsky, Shultz, & Takane, 1995). Two experiments were conducted to test these predictions using a concept learning task, measuring the number of trials required for participants to correctly classify 16 stimuli, consisting of key images defined by four binary dimensions, into two groups based on a relationship between the stimulus dimensions that was either simple or complex. The results indicated that the simple subtasking condition provides a good example of the effects of function compatibility, in that depending on what regularity participants extract in the first subset, learning of the rest of the task is facilitated or hindered as compared to all-at-once learning. The complex condition shows that some participants extracted a function in the first subset that was compatible with the rest of the task, thus facilitating learning as compared to all-at-once learning. However, for other participants, the knowledge acquired in subtasking is inaccessible in later learning. These findings provide evidence for the psychological validity of the simulations. Implications to part-whole transfer and applications in concept learning research are discussed.
| Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:QMM.24044 |
| Date | January 1996 |
| Creators | Varshney, Nicole Memorice. |
| Contributors | Shultz, Tom (advisor) |
| Publisher | McGill University |
| Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
| Language | English |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Format | application/pdf |
| Coverage | Master of Science (Department of Psychology.) |
| Rights | All items in eScholarship@McGill are protected by copyright with all rights reserved unless otherwise indicated. |
| Relation | alephsysno: 001541894, proquestno: MM19854, Theses scanned by UMI/ProQuest. |
Page generated in 0.0019 seconds